A Hybrid VMD-SVM Multi-Feature and Wavelet Modulus Maxima-Based Method for Fault Line Identification and Location in Distribution Networks with Single-Phase Ground Faults

Abstract

The transition from traditional power grids to new-type power systems increases the complexity of distribution networks, making fault currents weaker, more volatile, and richer in harmonics. To address the challenges of accurate identification, line selection, and location of single-phase ground faults in small-current grounding systems, an integrated method is proposed. A fault type identification step is incorporated prior to line selection and location. Sequence component characteristics are first used to identify the fault type. Multiple fault features are then constructed based on Variational Mode Decomposition (VMD) adaptively optimized by the Sparrow Search Algorithm (SSA). The Support Vector Machine (SVM) algorithm fuses these features for accurate fault line selection. Subsequently, the faulty phase is determined by analyzing the line-mode component characteristics of the selected fault line. Finally, a single-ended fault location method based on wavelet transform modulus maxima is applied to pinpoint the fault. A distribution network fault simulation model is established in Matlab/Simulink. Simulation results show that the proposed method achieves a fault-line selection accuracy of 98.4%, and the location error for a 100 km line is less than 1.673% The obtained accuracy and location precision demonstrate the effectiveness of the proposed VMD-SVM multi-feature fusion combined with wavelet modulus maxima ranging. Compared with conventional fuzzy theory methods, the SVM-based approach shows superior performance under limited sample conditions and complex fault scenarios. The proposed method effectively enhances the accuracy and reliability of single- phase ground-fault processing, offering a feasible technical solution for rapid isolation and precise location of faults in distribution networks.